Application of nonhomogenous Markov models for analyzing longitudinal caries risk.

OBJECTIVES: Markov modeling is a useful mathematical procedure for calculating probabilities of disease prognosis. Increasingly, Markov models are being applied in medical and health services research and also in social sciences research. The purpose of our study was to use the Markov process to determine time-dependent transition probabilities for caries-free children to convert to a caries-active state and to assess the impact of salivary mutans streptococci (MS) levels on caries status. METHODS: Our analysis was based on data obtained from a 6-year longitudinal study of risk factors associated with caries onset in children. RESULTS: Based on a two-state Markov model, the probability that a caries-free child would convert to a caries-active state during the study ranged between 0.0046 and 0.0471. The highest probability of converting from a caries-free state to a caries-active state was 0.0471 at age 8.5 years. CONCLUSIONS: In addition to standard statistical methods of analyzing longitudinal caries data, Markov models show promise for use in the analysis of caries risk.

Clinical outcomes for nursing caries treated using general anesthesia.

This study did follow-up examinations of 84 Medicaid-eligible children with nursing caries after they received treatment for nursing caries utilizing general anesthesia. Results at six months suggest that parents are unresponsive to follow-up care and over half of the children seen had new smooth surface caries lesions. While results are preliminary, they suggest that major changes are needed in the tertiary care of children with nursing caries.

Caries risk assessment in a longitudinal discrimination study.

Our objective was to develop and perfect a model for the assessment of risk of dental caries onset in children. Even though dental caries prevalence in children is continuing to decline, there is still a significant minority for whom it is a problem. In this study, we sought to ascertain whether a set of variables selected in a previous cross-sectional study could be used to differentiate between caries-free six-year-olds who would or would not subsequently present with clinically-detectable caries. A total of 472 caries-free six-year-olds--286 from a fluoridated community and 186 from a fluoride-deficient community--was selected. Clinical examinations for DMFS, dental fluorosis, and plaque were conducted. Stimulated whole saliva was collected for analysis of mutants streptococci, lactobacilli, total viable flora, and fluoride, calcium, and phosphate concentrations. A questionnaire was used for collection of demographic data as well as information on prior fluoride exposure, dietary habits, and oral hygiene practices. By means of linear discriminant analyses, it was possible to predict correctly which children would develop caries within six to 12 months (sensitivity) in 82.8% of cases and which children would not develop caries during that period (specificity) in 82.4% of cases.

Caries risk assessment by a cross-sectional discrimination model.

Although the prevalence of dental caries is continuing to decline, it still affects a majority of the US population and can be a serious problem for those afflicted. The objective of this project was to develop and perfect a model for assessment of risk of dental caries onset in children. In the first study, reported herein, a set of clinical, microbiological, biochemical, and socio-demographic variables was identified that distinguished, with an acceptable level of sensitivity and specificity, between children who had no previous caries experience and children who had high caries levels. A total of 313 children--age 12-15 years, 140 from a fluoridated community and 173 from a fluoride-deficient community--was selected on the basis of previous caries experience, either zero DMFS or high DMFS (> or = 6 in the fluoridated or > or = 8 in the fluoride-deficient community). Clinical exams for DMFS, dental fluorosis, and plaque were conducted. Stimulated whole saliva was collected for analysis of mutans streptococci, lactobacilli, total viable flora, and fluoride concentration. A questionnaire was used for collection of demographic data as well as information on prior fluoride exposure, dietary habits, and oral hygiene practices. By means of discriminant analyses, with use of seven key clinical and laboratory variables, it was possible for zero-DMFS subjects to e classified correctly (specificity) in 77.6% of cases in the fluoridated community and in 86.1% of cases in the fluoride-deficient community. High-caries subjects were classified as such (sensitivity) in 79.3% and 88.1% of cases, respectively.

Assessment of the effect of chemotherapy and radiotherapy on the auditory function of children with cancer.

Forty children with malignant disease underwent full audiological assessment with pure tone audiometry, tympanometry and measurement of brain stem auditory evoked potentials in an attempt to detect auditory dysfunction resulting from treatment. Significant auditory dysfunction was observed in 8 children. Three children with evidence of a conductive or sensorineural deafness on pure tone audiometry had normal brain stem evoked potentials. One child with a high tone hearing loss and 4 children with no detectable hearing loss on audiometry showed significant differences in the interaural interpeak latencies of selected brain stem auditory evoked potentials. In 2 children these abnormalities had resolved on subsequent testing, 2 had evidence of persistent retrocochlear dysfunction and the remaining child died prior to further evaluation. Routine measurement of brain stem auditory evoked potentials is probably not necessary for the majority of children undergoing treatment for cancer, although such measurements are of value when differentiation between cochlear and retrocochlear damage is required.